Switch cabinet hidden danger prediction method and device based on multi-physics field fusion, storage medium and equipment

Abstract

The invention discloses a switch cabinet hidden danger prediction method and device based on multi-physics field fusion, a storage medium and equipment. The switch cabinet hidden danger prediction method comprises the steps of performing discretization processing on acquired operation data of switch cabinet equipment; inputting the discretized operation data into a Bayesian network model to obtain a prediction result of the equipment hidden danger; performing degradation processing on the prediction result to obtain degradation data; and fusing the degradation data and the multi-physical field simulation data of the switch cabinet equipment, and carrying out hazard reasoning drilling on equipment hidden dangers based on the fused data to obtain a reasoning result. Real-time data and a physical mechanism can be effectively fused, and accurate prediction can be performed under a small sample condition.

Description

Technical Field

[0001] This invention belongs to the field of intelligent control system technology, specifically, it relates to a method, device, storage medium and equipment for predicting potential hazards in switchgear based on multi-physics field fusion. Background Technology

[0002] Switchgear is a key piece of equipment in power systems used for power distribution, control, and protection; its operating status directly affects the safety and stability of the power grid. Therefore, early prediction and hazard assessment of potential hidden dangers within switchgear are of great significance for implementing condition-based maintenance and preventing faults.

[0003] Currently, the reasoning and prediction methods for potential hazards in switchgear equipment are mainly based on the following types of technologies: The first type is machine learning methods, which rely on manual feature engineering and shallow learning models, such as support vector machines or gradient boosting decision trees, to fit and predict data patterns; the second type is deep learning methods, which use deep neural networks to automatically extract features and make inferences, such as convolutional long short-term memory networks or Transformer models; the third type is expert knowledge system methods, which build rule bases based on the experience of domain experts and make judgments through logical reasoning; the fourth type is probabilistic graphical methods, such as Bayesian networks, which perform uncertainty reasoning by modeling the probabilistic dependencies between variables.

[0004] However, these existing technologies all have significant limitations when applied to predicting early-stage hazards in switchgear. First, machine learning and deep learning methods heavily rely on a large number of high-quality historical fault samples for model training. However, early-stage hazard symptoms in switchgear are weak and samples are scarce, making data-driven models prone to overfitting and exhibiting poor generalization ability. Furthermore, their black-box nature means that the prediction results lack physical mechanism support, making them difficult for maintenance personnel to trust. Second, while expert knowledge systems are interpretable, their rules are often rigid, making it difficult to adapt to complex changes in field operating conditions, and they cannot quantify the uncertainty of prediction results. Finally, and more critically, existing methods mostly start from historical data or static rules in isolation, failing to deeply integrate real-time monitoring data with simulation data reflecting the multi-physics state inside the equipment. This results in prediction models that cannot accurately capture the early fault evolution process caused by the coupling of multi-physics fields, making it difficult to reliably extrapolate and verify the development trend of hazards.

[0005] Therefore, there is an urgent need in this field for a new method that can effectively integrate real-time data and physical mechanisms, and can intelligently predict and reason about potential hazards in switchgear under small sample conditions. Summary of the Invention

[0006] (I) The technical problem to be solved by the present invention

[0007] How can we provide a method for predicting potential hazards in switchgear based on multi-physics field fusion that can effectively integrate real-time data and physical mechanisms and make predictions under small sample conditions?

[0008] (II) Technical Solution Adopted in this Invention

[0009] A method for predicting potential hazards in switchgear based on multi-physics fusion, the method comprising:

[0010] The acquired operating data of the switchgear equipment is discretized.

[0011] The discretized operating data is input into a Bayesian network model to obtain the prediction results of potential equipment hazards;

[0012] The prediction results are then subjected to degradation processing to obtain degraded data;

[0013] The degradation data and multiphysics simulation data of the switchgear equipment are fused together, and the hazard reasoning exercise of potential equipment hazards is carried out based on the fused data to obtain the reasoning results.

[0014] Optionally, the acquired operating data of the switchgear equipment is discretized, including:

[0015] The operating data is divided into intervals, and the operating data includes current, temperature, voltage and signal quantity.

[0016] Optionally, the discretized operational data is input into a Bayesian network model to obtain prediction results of potential equipment hazards, including:

[0017] Obtain the trained Bayesian network model;

[0018] The discretized operational data is input into the trained Bayesian network model to obtain the types of equipment hazards and their probabilities of occurrence.

[0019] Optionally, the prediction results are subjected to degradation processing to obtain degradation data, including:

[0020] The predicted results are subjected to simulated degradation processing to obtain degradation data, which includes temperature, current, load and voltage parameters.

[0021] Optionally, the multiphysics simulation data of the switchgear equipment includes field distribution data obtained by performing coupled simulation calculations of electromagnetic field, thermal field and structural stress field of the switchgear based on finite element analysis software.

[0022] Optionally, based on the fused data, a hazard reasoning exercise for potential equipment problems is conducted to obtain the reasoning results, including:

[0023] The Monte Carlo method is used to perform uncertainty sampling on the fused data to obtain sampled data;

[0024] The surrogate model is used to perform inference and prediction on the sampled data to obtain the inference results;

[0025] The inference results were verified by multiphysics coupling simulation.

[0026] Optionally, the Monte Carlo method includes Markov chain Monte Carlo sampling and Latin hypercube sampling.

[0027] This application also discloses a switchgear hazard prediction device based on multiphysics field fusion, the switchgear hazard prediction device comprising:

[0028] The data processing module is configured to: discretize the acquired operating data of the switchgear equipment;

[0029] The prediction module is configured to input the discretized operating data into a Bayesian network model to obtain the prediction results of potential equipment hazards.

[0030] A degradation processing module is configured to: perform degradation processing on the prediction results to obtain degradation data;

[0031] The fusion reasoning module is configured to fuse the degradation data and the multiphysics simulation data of the switchgear equipment, perform hazard reasoning exercises based on the fused data, and obtain reasoning results.

[0032] This application also discloses a computer-readable storage medium storing a switchgear hazard prediction program based on multi-physics field fusion. When the switchgear hazard prediction program based on multi-physics field fusion is executed by a processor, it implements the above-mentioned switchgear hazard prediction method based on multi-physics field fusion.

[0033] This application also discloses a computer device, which includes a computer-readable storage medium, a processor, and a multi-physics fusion-based switchgear hazard prediction program stored in the computer-readable storage medium. When the multi-physics fusion-based switchgear hazard prediction program is executed by the processor, it implements the above-mentioned multi-physics fusion-based switchgear hazard prediction method.

[0034] (III) Beneficial Effects

[0035] This invention discloses a method, apparatus, storage medium, and device for predicting potential hazards in switchgear based on multiphysics field fusion, which has the following technical advantages compared to existing methods:

[0036] 1. In this solution, a Bayesian network model is constructed based on discretized data to predict equipment hazard and to fuse deteriorated operational data with multiphysics simulation data; a dual-driven intelligent prediction framework is built. It does not simply rely on historical statistical patterns, nor does it completely depend on fixed expert rules. Instead, it utilizes the Bayesian network, a probabilistic graphical model, to deeply fuse various measurement point data of the equipment with multiphysics simulation data. Bayesian networks excel at expressing probabilistic dependencies between variables and can effectively combine equipment mechanisms with uncertainties in operational data for reasoning; early-stage hazard samples for switchgear are extremely limited, and traditional data-driven methods are prone to overfitting or failure.

[0037] Bayesian networks, with their probabilistic inference characteristics, are naturally suited for handling scenarios with small samples and high uncertainty. They do not pursue absolutely accurate predictions, but rather provide the probability of potential hazards occurring, offering a more scientific basis for decision-making and greatly improving the feasibility of predictions in situations with scarce samples. By fusing multiphysics simulation data, the prediction model is no longer a "black box." Decision-makers can not only know what potential hazards might occur, but also understand why they might occur from a physical mechanism perspective. This significantly enhances the credibility of the prediction results, making them easier for domain experts to accept and adopt. Compared to fixed expert systems, this data-mechanism fusion model can continuously update the probability parameters in the network with new data, adapting to different operating conditions and different types of switchgear equipment, reducing the cost and difficulty of model maintenance and migration.

[0038] 2. In this solution, the predicted results are degraded to generate degraded operational data. Degradation processing is introduced to simulate the deterioration of the equipment's current operating parameters in a virtual space, generating a series of "virtual operational data" representing the continuous deterioration of the equipment's condition. This data serves as the starting point for subsequent hazard reasoning exercises. In reality, it is extremely difficult to obtain complete data on the entire process of equipment development from a potential hazard to a serious failure, making it impossible to train models for hazard simulation. This method, by artificially constructing a fault evolution sequence, perfectly solves the core pain point of lacking training samples in the hazard reasoning stage, making it possible to predict the future development trend of potential hazards. It simulates what kind of serious failure might evolve into and what impact it would have if the current hazard were not addressed. This shifts decision-making from post-event remediation or in-event emergency response to pre-event prevention. It allows for the assessment of the potential risk levels of different hazards, thereby optimizing maintenance strategies and resource allocation. All simulations are conducted in a digital virtual environment, without any dangerous operations on the actual equipment, allowing for zero-risk and zero-cost testing of various extreme situations. This provides a powerful digital tool for developing safe and effective emergency plans.

[0039] 3. In this scheme, Monte Carlo methods are used for uncertainty sampling and surrogate model construction to improve inference efficiency. When performing verification using multiphysics simulations, direct high-fidelity coupled simulation calculations are extremely costly and time-consuming, failing to meet the needs of real-time or near-real-time prediction. This scheme introduces Markov chain Monte Carlo and Latin hypercube sampling to efficiently explore the parameter space and constructs a surrogate model based on the sampling points to approximate the complex physical simulation model. The computation speed of the surrogate model is several orders of magnitude faster than high-precision physical simulation, enabling simulation calculations that originally took hours or even days to be completed in seconds or minutes. This achieves near real-time multiphysics hazard deduction and verification, making this technology feasible in practical engineering applications. Through advanced sampling methods such as MCMC and LHS, the possible distributions of parameters can be covered to the maximum extent with fewer sampling points, ensuring that the constructed surrogate model is both efficient and fully reflects the inherent uncertainty of the physical system, making the deduction conclusions more reliable and comprehensive in a statistical sense. This method does not completely abandon high-precision simulation but uses it to train the surrogate model, ultimately using a lightweight surrogate model in the deduction phase. This cleverly balances the trade-off between computational accuracy and speed, providing an efficient technical approach for digital twins and real-time simulation of complex industrial equipment. Attached Figure Description

[0040] Figure 1 This is a flowchart of a switchgear hazard prediction method based on multiphysics field fusion according to one or more embodiments.

[0041] Figure 2 This is a schematic diagram of a switchgear hazard prediction method device based on multiphysics field fusion according to one or more embodiments.

[0042] Figure 3 This is a schematic diagram of a computer device according to one or more embodiments. Detailed Implementation

[0043] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0044] Before describing the various embodiments of this application in detail, the technical concept of this application is first briefly described: Current methods for predicting potential hazards in switchgear either rely on a large amount of sample data to train the model or struggle to make accurate predictions based on real-time data. Therefore, this application provides a method for predicting potential hazards in switchgear based on multiphysics field fusion. The key improvement lies in using a Bayesian network model to predict the operating data of the switchgear equipment, obtaining preliminary prediction results. After deteriorating the prediction results, the deteriorated data and multiphysics field simulation data are fused and hazard reasoning exercises are performed to obtain the reasoning results. The specific principles of the switchgear hazard prediction method based on multiphysics field fusion of this application are described below with reference to more embodiments.

[0045] Specifically, such as Figure 1 As shown, the switchgear hazard prediction method based on multiphysics field fusion in this embodiment includes the following steps:

[0046] Step S10: Discretize the obtained operating data of the switchgear equipment;

[0047] Step S20: Input the discretized operating data into the Bayesian network model to obtain the prediction results of equipment hidden dangers;

[0048] Step S30: Degrade the prediction results to obtain degraded data;

[0049] Step S40: Fuse the deterioration data and the multi-physics simulation data of the switchgear equipment, and conduct a hazard reasoning exercise for potential equipment hazards based on the fused data to obtain the reasoning results.

[0050] In one or more embodiments, during the data acquisition step, historical and real-time operating data of the switchgear equipment are acquired. The operating data includes current, voltage, temperature, and signal quantities. These signal quantities include active power, reactive power, average partial discharge, peak partial discharge, opening and closing signals, and protection action signals. Equipment ledger information is also acquired simultaneously. This ensures that the input information for the hazard prediction model is comprehensive and multi-dimensional, covering multiple key state parameters such as electrical, mechanical, and insulation aspects, providing a sufficient data foundation for accurate prediction.

[0051] In one or more embodiments, the acquired operating data of the switchgear equipment is discretized by dividing the operating data into intervals. For example, continuous measurement point data is divided into predefined intervals; wherein, the current is divided into multiple intervals including (1490-1980), (2290-2490), and (2600-2850); the temperature is divided into multiple intervals including (58-67), (67-75), (75-83), and (83-95); the voltage is divided into multiple levels including 75%, 80%, 90%, 110%, and 130% as a percentage of the rated voltage; and the signal quantity is converted to a Boolean state. By dividing the current, temperature, voltage, and signal quantity into intervals, the processing efficiency of the original data and the interpretability of the model are improved, making continuous data more suitable for probabilistic inference models.

[0052] In one or more embodiments, the discretized operating data is input into a Bayesian network model to obtain the prediction result of equipment hazards, including: obtaining a trained Bayesian network model; inputting the discretized operating data into the trained Bayesian network model to obtain the type of equipment hazard and its probability of occurrence.

[0053] The training samples required to construct the Bayesian network model are generated by modifying equipment defect data. The modification strategy is as follows: based on the importance of defects in the equipment defect data, common minor defects are defined as potential hazard samples, and major defects are defined as equipment defect samples. The discretized data is used as input variables; the output is the type of equipment hazard and its probability of occurrence. This transforms abstract equipment defects into quantifiable model training samples and outputs concrete hazard types and their probabilities of occurrence, making the prediction results more intuitive and practical. Hazard types include poor contact.

[0054] In one or more embodiments, the predicted results are subjected to simulated degradation processing to obtain degradation data, which includes temperature, current, load, and voltage parameters. Exemplarily, in the simulated degradation processing step, the parameters are adjusted for deterioration based on their original values; for example, the phase A temperature is degraded from its original range to (68–73°C), the phase A current to (485–500A), the load to (5–7MW), and the voltage to (7–8kV). By performing simulated degradation operations on temperature, current, load, and voltage parameters, the equipment condition deterioration path can be safely and cost-effectively simulated in a virtual environment, providing crucial data input for subsequent hazard reasoning exercises.

[0055] In one or more embodiments, the multiphysics simulation data of the switchgear equipment includes field distribution data obtained by performing coupled simulation calculations of electromagnetic field, thermal field, and structural stress field of the switchgear based on finite element analysis software. By using multiphysics simulation data, including coupled simulation results of electromagnetic field, thermal field, and structural field, the limitations of a single data dimension are overcome. This allows for a deeper understanding of the intrinsic mechanisms and potential impacts of hidden dangers from multiple physical levels, including electromagnetic, thermal, and structural aspects, thereby improving the accuracy and completeness of the reasoning.

[0056] In one or more embodiments, degradation data and multiphysics simulation data of switchgear equipment are fused, and hazard reasoning exercises for potential equipment hazards are performed based on the fused data to obtain reasoning results, including:

[0057] Step S401: Uncertainty sampling is performed on the fused data using the Monte Carlo method to obtain sampled data. The Monte Carlo method includes Markov chain Monte Carlo sampling and Latin hypercube sampling, which combines the advantages of different sampling methods, efficiently covers the parameter space and characterizes its probability distribution, laying a solid foundation for building a high-precision surrogate model.

[0058] Step S402: Use a surrogate model to perform inference and prediction on the sampled data to obtain inference results. A Latin hypercube sampling method is used to extract sample points from the multiphysics simulation results to construct a surrogate model, thereby improving inference efficiency. The constructed surrogate model replaces the computationally expensive full-process coupled simulation for efficient inference trend judgment and verification. The surrogate model is constructed using Gaussian process regression or a neural network algorithm to approximate the complex nonlinear relationship between input parameters and multiphysics output.

[0059] Step S403: Verify the inference results through multiphysics coupling simulation. Performing multiphysics coupling simulation to verify the implementation of the inference results significantly improves the computational efficiency and engineering feasibility of hazard inference exercises, and ensures that the inference results fully consider the uncertainties of the system, resulting in more rigorous conclusions.

[0060] In one or more embodiments, the switchgear hazard prediction method based on multi-physics field fusion further includes a time-series prediction step: based on time-series data analysis methods, the operating status trend of the equipment from the current time T to the future time T+1 is predicted; by performing time-series trend analysis on the equipment hazard prediction results, the data change trend from time T to time T+1 is predicted, providing a forward-looking prediction capability for the development trend of equipment status, supporting maintenance personnel to make proactive interventions and forward-looking decisions.

[0061] In one or more embodiments, the switchgear hazard prediction method based on multiphysics fusion further includes visual rendering of the reasoning and simulation process to demonstrate the fault simulation process. Specifically, the visualization rendering steps include: dynamically rendering and graphically displaying the reasoning and simulation process and results of equipment hazard hazards, including fault simulation process animations and data trend comparison charts. By visually rendering the reasoning and simulation process to demonstrate the fault simulation process, the understandability of the reasoning process and results is greatly improved, making it easier for maintenance personnel to quickly locate problems, understand the fault evolution process, and formulate measures.

[0062] The working principle of this multi-physics-based switchgear hazard prediction method is as follows: First, multi-dimensional real-time operating data of the switchgear is collected, and continuous measurement data such as current, temperature, and voltage are preprocessed by discretizing intervals based on business rules, converting them into an input form suitable for probabilistic model processing. Then, a Bayesian network model is trained using samples generated by modifying defect data. Through its probabilistic reasoning ability, the measurement point data and equipment mechanism are fused to output the early hazard type and probability of occurrence. To further evaluate the development trend of hazards, the system simulates the degradation of key parameters of the predicted state, generates inference input data characterizing the deterioration of equipment condition, and fuses this data with the physical field data obtained by electromagnetic-thermal-structural multi-physics coupling simulation. Based on this fused data, Markov chain Monte Carlo and Latin hypercube sampling methods are used to handle uncertainty and construct a surrogate model to replace the computationally expensive full simulation, ultimately achieving efficient and visualized equipment hazard inference exercise and trend verification.

[0063] like Figure 2 As shown in Embodiment 2 of this application, a method and apparatus for predicting potential hazards in switchgear based on multiphysics field fusion is also disclosed. The switchgear hazard prediction apparatus includes a data processing module 100, a prediction module 200, a degradation processing module 300, and a fusion reasoning module 400. The data processing module 100 is configured to discretize the acquired operating data of the switchgear equipment. The prediction module 200 is configured to input the discretized operating data into a Bayesian network model to obtain the prediction result of the equipment hazard. The degradation processing module 300 is configured to degrade the prediction result to obtain degradation data. The fusion reasoning module 400 is configured to fuse the degradation data and the multiphysics field simulation data of the switchgear equipment, and perform hazard reasoning exercises based on the fused data to obtain the reasoning result. The detailed processing procedures of each module are described in Embodiment 1 and will not be repeated here.

[0064] In one or more embodiments, a computer-readable storage medium stores a program for a switchgear hazard prediction method based on multiphysics field fusion, which, when executed by a processor, implements the aforementioned switchgear hazard prediction method based on multiphysics field fusion.

[0065] In one or more embodiments, such as Figure 3 As shown, the computer device includes a processor 12, an internal bus 13, a network interface 14, and a computer-readable storage medium 11. The processor 12 reads the corresponding computer program from the computer-readable storage medium and runs it, forming a request processing device at the logical level. Of course, in addition to the software implementation, one or more embodiments of this specification do not exclude other implementation methods, such as logic devices or a combination of hardware and software, etc. That is to say, the execution subject of the following processing flow is not limited to each logic unit, but can also be hardware or logic devices. The computer-readable storage medium 11 stores a program for a switchgear hazard prediction method based on multi-physics field fusion. When the processor executes the program for the switchgear hazard prediction method based on multi-physics field fusion, it implements the above-mentioned switchgear hazard prediction method based on multi-physics field fusion.

[0066] Computer-readable storage media, including both permanent and non-permanent, removable and non-removable media, can store information using any method or technology. The information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer-readable storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage, quantum memory, graphene-based storage media or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0067] To demonstrate the advantages of the switchgear hazard prediction method based on multi-physics field fusion, staff conducted a hazard intelligent prediction and reasoning exercise on a Siemens 8DA10 10kV metal-clad central switchgear (numbered KG-12) at the operation and maintenance center of a 220kV smart substation in a certain city's power grid on May 10, 2024.

[0068] First, by using current and voltage data collected by a PT100 platinum resistance temperature sensor and an ABB ETU760 microprocessor protection device deployed within the cabinet, as well as partial discharge signals monitored by a Zhengzhou Dafang DT series ultrasonic partial discharge sensor, multi-dimensional operational data such as the A-phase contact temperature of 72.3 degrees Celsius, load current of 2380 amps, operating voltage of 10.2 kV, and partial discharge peak value of 45 dB / µV are acquired in real time. After this data is uploaded to the cloud platform via the Huawei Atlas500 edge computing gateway, the system immediately calls the preprocessing module to discretize the current to the 2290-2490 amp range, classify the temperature to the 67-75 degree Celsius range, and convert the voltage to 102% of the rated value according to preset rules.

[0069] The preprocessed data was input into a Bayesian network prediction model built on the Netica platform. This model, trained on 300 historical defect records, inferred that the probability of a poor contact defect in phase A of the cabinet was 87.2%. To assess the consequences of this defect, the system activated a degradation simulation engine, manually setting the phase A temperature to 73 degrees Celsius, increasing the current to 495 amperes, increasing the load to 6.8 megawatts, and decreasing the voltage to 7.6 kilovolts, generating an extreme operating condition input dataset.

[0070] Subsequently, the ANSYS Workbench multiphysics simulation platform was invoked, and electromagnetic-thermal structural coupling calculations were performed using the degraded parameters as boundary conditions. This yielded physical field data including a maximum temperature of 162 degrees Celsius in the contact area, a thermal stress distribution cloud map, and a deformation displacement of 0.38 mm. To accelerate the verification process, the system employed Latin hypercube sampling technology to extract 500 sample points from the simulation results and constructed a surrogate model based on a Gaussian process regression algorithm, replacing the time-consuming full finite element analysis.

[0071] Ultimately, through 10,000 iterations of Monte Carlo simulation, dynamic visualization showed that the probability of this contact hazard developing into a permanent welding failure after 48 hours exceeded 65%, while also rendering a 3D simulation animation of arc burning the cabinet. Based on this simulation conclusion, maintenance personnel immediately arranged preventative maintenance such as live cleaning and bolt tightening, successfully averting a potential switchgear explosion accident.

[0072] The specific embodiments of the present invention have been described in detail above. Although some embodiments have been shown and described, those skilled in the art should understand that modifications and improvements can be made to these embodiments without departing from the principles and spirit of the present invention as defined by the claims and their equivalents, and such modifications and improvements should also be within the protection scope of the present invention.

Claims

1. A method for predicting potential hazards in switchgear based on multiphysics field fusion, characterized in that, The method for predicting potential hazards in switchgear includes: The acquired operating data of the switchgear equipment is discretized. The discretized operating data is input into a Bayesian network model to obtain the prediction results of potential equipment hazards; The prediction results are then subjected to degradation processing to obtain degraded data; The degradation data and multiphysics simulation data of the switchgear equipment are fused together, and the hazard reasoning exercise of potential equipment hazards is carried out based on the fused data to obtain the reasoning results.

2. The method for predicting potential hazards in switchgear based on multiphysics field fusion according to claim 1, characterized in that, The acquired operating data of the switchgear equipment is discretized, including: The operating data is divided into intervals, and the operating data includes current, temperature, voltage and signal quantity.

3. The method for predicting potential hazards in switchgear based on multiphysics field fusion according to claim 1, characterized in that, The discretized operational data is input into a Bayesian network model to obtain predictions of potential equipment hazards, including: Obtain the trained Bayesian network model; The discretized operational data is input into the trained Bayesian network model to obtain the types of equipment hazards and their probabilities of occurrence.

4. The method for predicting potential hazards in switchgear based on multiphysics field fusion according to claim 1, characterized in that, The prediction results are subjected to degradation processing to obtain degradation data, including: The predicted results are subjected to simulated degradation processing to obtain degradation data, which includes temperature, current, load and voltage parameters.

5. The method for predicting potential hazards in switchgear based on multiphysics field fusion according to claim 1, characterized in that, The multiphysics simulation data of the switchgear equipment includes field distribution data obtained by coupled simulation calculation of electromagnetic field, thermal field and structural stress field of the switchgear based on finite element analysis software.

6. The method for predicting potential hazards in switchgear based on multiphysics field fusion according to claim 1, characterized in that, Based on the fused data, a hazard reasoning exercise for potential equipment problems was conducted, yielding the following reasoning results: The Monte Carlo method is used to perform uncertainty sampling on the fused data to obtain sampled data; The surrogate model is used to perform inference and prediction on the sampled data to obtain the inference results; The inference results were verified by multiphysics coupling simulation.

7. The method for predicting potential hazards in switchgear based on multiphysics field fusion according to claim 1, characterized in that, The Monte Carlo method includes Markov chain Monte Carlo sampling and Latin hypercube sampling.

8. A switchgear hazard prediction device based on multiphysics field fusion, characterized in that, The switchgear hazard prediction device includes: The data processing module is configured to: discretize the acquired operating data of the switchgear equipment; The prediction module is configured to input the discretized operating data into a Bayesian network model to obtain the prediction results of potential equipment hazards. A degradation processing module is configured to: perform degradation processing on the prediction results to obtain degradation data; The fusion reasoning module is configured to fuse the degradation data and the multiphysics simulation data of the switchgear equipment, perform hazard reasoning exercises based on the fused data, and obtain reasoning results.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a switchgear hazard prediction program based on multi-physics field fusion. When the switchgear hazard prediction program based on multi-physics field fusion is executed by a processor, it implements the switchgear hazard prediction method based on multi-physics field fusion as described in any one of claims 1 to 7.

10. A computer device, characterized in that, The computer device includes a computer-readable storage medium, a processor, and a program for predicting potential hazards in switchgear based on multiphysics fusion stored in the computer-readable storage medium. When the program for predicting potential hazards in switchgear based on multiphysics fusion is executed by the processor, it implements the method for predicting potential hazards in switchgear based on multiphysics fusion as described in any one of claims 1 to 7.

Images